MetaMCP

MetaMCP is an MCP server that operates as a reverse proxy and aggregator, combining multiple downstream Model Context Protocol servers into a single unified endpoint. It connects to the MetaMCP application to retrieve configurations for tools, prompts, and resources, dynamically fetching capability lists from child servers and exposing them as an aggregated interface to clients such as Claude Desktop. Software developers and AI tool builders use MetaMCP to prevent configuration bloat and manage tool namespaces across disparate services. By handling request routing behind a single client connection, it directs specific tool execution calls directly to the responsible underlying MCP process. MetaMCP supports multiple transport mechanisms, including standard input/output (stdio), Server-Sent Events (SSE), and Streamable HTTP, alongside Docker host resolution configurations. Users can toggle workspace profiles via the companion application to update active toolsets without manually rewriting their MCP client settings.

Category: AI & LLM Tooling

Tags: gateway, orchestration, proxy, routing

Visit MetaMCP

How to install and configure MetaMCP

1. Obtain your MetaMCP API key from your MetaMCP App configuration. 2. Configure your MCP client (such as Claude Desktop) using the npx command. 3. Add the following entry to your mcpServers configuration file: json { "mcpServers": { "MetaMCP": { "command": "npx", "args": ["-y", "@metamcp/mcp-server-metamcp@latest"], "env": { "METAMCP_API_KEY": "<your api key>" } } } } 4. Alternatively, run the server manually across different transports like SSE using mcp-server-metamcp --metamcp-api-key <your-api-key> --transport sse --port 12006 or install via Smithery using npx -y @smithery/cli install @metatool-ai/mcp-server-metamcp --client claude.

What you can do with MetaMCP

• Aggregating multiple local and remote MCP servers into a single unified configuration endpoint for Claude Desktop. - Dynamically swapping tool and prompt sets across project workspaces without editing configuration files manually. - Routing tool execution calls across namespace-isolated child MCP processes automatically based on client requests. - Running MCP tooling in containerized environments by mapping localhost endpoints to host Docker internal addresses. - Bridging stdio-based downstream servers to network clients using Server-Sent Events or Streamable HTTP transports.

Key facts

• https://github.com/metatool-ai/mcp-server-metamcp
• AI & LLM Tooling
• gateway, orchestration, proxy, routing

Part of MCP Servers

How do I install MetaMCP?

MetaMCP can be configured in your MCP client by adding the npm execution command to your configuration file, passing your MetaMCP API key in the environment variables. Alternatively, you can use the Smithery CLI command with the claude flag or install the package directly via npm to run it from your command line.

What transports are supported by MetaMCP?

MetaMCP supports stdio as its default transport, along with Server-Sent Events and Streamable HTTP. Network transports can be run by specifying the transport and port flags during execution, with optional support for API authentication and stateless modes.

Which MCP clients work with MetaMCP?

MetaMCP is designed to be compatible with any client that follows the Model Context Protocol, including Claude Desktop, Cursor, and Windsurf, since it acts as a transparent reverse proxy aggregating underlying servers.

Is MetaMCP still actively maintained?

The local proxy package repository has a deprecation notice indicating that its functionality has shifted into the MetaMCP 2.0 all-in-one architecture, hosted in the primary MetaMCP repository.